Based on Particle Swarm Intelligence Algorithm Optimized Extreme Learning Machine for Vibration Prediction of Smoke Machine Bearings

The performance prediction of vibration data for tobacco machinery rolling bearings is of great significance for online fault pre evaluation and diagnosis. This article proposes a particle swarm optimization algorithm for predicting rolling bearing vibration using extreme learning machines. Firstly, phase space reconstruction is performed on the rolling bearing vibration time series data. Due to the non-linear functional relationship between the input parameters and the output responses in phase space, this paper adopts particle swarm optimization algorithm to optimize the input weights and thresholds of extreme learning machine, and uses the optimized extreme learning machine to establish a relationship model between input and output in phase space of rolling bearing vibration data sequence. According to the dynamic equivalence between the reconstructed phase space and the original system in the sense of homeomorphism, the phase space prediction model can be used to predict the time series data of the original rolling bearing after inversion. The method proposed in this article is used for predicting and verifying vibration data of rolling bearings, and the results show that the extreme learning machine optimized by particle swarm optimization has better predictive performance than the pure extreme learning machine.